馃攳
DDR5 Scalping is SOLVED. - Just buy DDR4 - YouTube
Channel: unknown
[0]
- So you've come looking for DDR5?
[3]
I'll cut you a deal. Only
double retail pricing.
[7]
It's a steal, for me, that is.
[11]
(man laughing)
[18]
- Scalpers have taken the
fledgling DDR5 market hostage.
[21]
Look around you.
[22]
Everything on Neweeg is
out of stock right now.
[24]
And low end 16 gig modules
of DDR5 are being sold
[28]
for as much as 32 gigs of
Trident Z is supposed to be.
[33]
Meanwhile on eBay,
[35]
you're looking at anywhere
from 700 to $1,000
[38]
for a 32 gig kit of the good stuff.
[41]
We're in a bad place right
now, but there's a light,
[44]
a light of justice because
the fact that DDR5 is newer,
[48]
shinier doesn't mean
that you need to buy it.
[51]
Unlike our sponsor,
GlassWire lets you see past
[53]
and present network activity,
detect malware on your PC
[56]
or Android device, and
block its connections
[58]
to prevent things from getting worse.
[60]
Use offer code Linus to get
25% off at the link down below.
[64]
(upbeat music)
[72]
I know, I know.
[74]
We did a whole video about
why DDR5 is super cool,
[78]
and it really is,
[80]
but it's got a lot working
against it right now too.
[83]
You've got first gen RAM chips interfacing
[85]
with first gen memory controllers
[87]
over imperfectly optimized
first gen trace layouts.
[90]
And it's perfectly natural to
expect all this first genery
[95]
to take a while to reach
its full potential.
[97]
I mean think about how the
first PlayStation 2 games
[99]
ended up looking worse than
the last PlayStation 1 games
[102]
because developers had
no idea yet how to unlock
[105]
the full potential of the Emotion Engine.
[107]
And even in the computer memory space,
[109]
this is nothing new.
[111]
DDR2, DDR3, DDR4.
[113]
If you look at contemporaneous reviews,
[115]
you'll find a lot of reports saying
[117]
that there is no practical
benefit to the upgrade
[120]
and that it is certainly not
worth the early adopter tax.
[124]
Like look at this dork back in 2014,
[127]
the synthetic benchmark graphs show DDR4
[130]
absolutely wiping the floor
with DDR3 in raw bandwidth.
[134]
But when we look at the numbers
in real-world applications,
[137]
even this relatively low spec kit of DDR3,
[141]
beats out the fastest DDR4
that was available at the time.
[145]
Of course, as we all know, DDR4,
[148]
like its predecessors, ended
up getting the last laugh
[151]
with mid-generation refinements
[153]
that led to much higher performance
[155]
at ultimately lower costs.
[157]
I mean, the hot item back
when that video was shot was,
[160]
3000 mega transfers per second CL16.
[164]
Nowadays, we would call
that budget at best,
[167]
cheapo at worst.
[169]
And the same thing will
eventually happen with DDR5,
[173]
just not today.
[174]
Unlike our new water bottles,
[176]
those are already proven superior
[178]
and you can get 'em right now,
[179]
no scalping required, lttstore.com.
[181]
To illustrate our point,
we ran a bevy of benchmarks
[183]
using our Core I9 12900K
paired with an RTX 3090.
[187]
This is admittedly a
very high spec machine
[190]
that not everyone will be using,
[192]
but it allows both of
our memory technologies
[194]
to put their best foot forward
[196]
by removing system bottlenecks elsewhere.
[198]
For DDR5, we've included both,
[200]
the highest currently available JNX speed
[202]
and a 5,200 mega transfer per second kit,
[204]
which was the fastest that our
system could handle reliably.
[208]
And then we've included
a handful of common
[210]
DDR4 configurations.
[212]
Kicking things off, our
high end DDR5 pulls off
[214]
a respectable 5% lead over our
fastest DDR4 kit in F1 2021.
[219]
Though it should be noted
that we're having issues
[222]
running more than two sticks of the stuff
[223]
at a time right now.
[224]
And the other problem is
that our lower end DDR5 kit
[228]
drops off significantly
from its bigger brother.
[231]
In fact, while the average
FPS drop looks fine,
[234]
it trails even our top
spec DDR4 by as much as 10%
[239]
in minimum frame times.
[240]
And that could easily
manifest as little hitches
[243]
or stutters in your gameplay.
[245]
Forza is a little bit kinder to our DDR5,
[247]
but the improvement
offered by the high end kit
[249]
is negligible and the
benefit over DDR4 is under 5%
[253]
in both cases.
[254]
Flight Simulator CPU
intensive nature seems like
[256]
it should be a good fit for DDR5, but no.
[260]
Both kits are actually
consistently between one
[262]
and 4% slower than our fast DDR4 kit.
[266]
That's surprising, but
not as surprising as CS:GO
[269]
another game that CPU heavy.
[272]
While the high-end kit
stays roughly equivalent
[274]
with a few frames gained or
lost, depending on the metric,
[277]
our lower end kit is on
par or worse than DDR4,
[281]
running 400 mega trumps
frames per second lower.
[284]
Civ VI for its part was more
or less the same regardless
[287]
with less than half a
second separating our DDR5
[289]
from our fastest DDR4.
[291]
Taken together then the
gaming tests average out to
[294]
about a 2% win for the high-end kit
[296]
and a 2% loss for the low-end
kit compared to high end DDR4,
[300]
which, and I cannot emphasize
this enough costs less
[304]
than half as much at retail.
[307]
And that is pre scalper markup.
[309]
Perhaps productivity will produce
[311]
DDR5s proper performance potential.
[314]
Cinebench sure doesn't think so.
[315]
It's a total wash with a
measly 1% difference or less
[319]
between our fastest DDR4 and DDR5 kits.
[323]
Compiling Firefox though,
gives DDR5 its first major win
[326]
even the lower end kit
picking up a 6% faster time
[330]
compared to our best DDR4
[332]
and the higher end kit being
nearly a minute faster.
[335]
Code compilation does hit system
memory pretty hard though.
[338]
So this isn't a huge surprise,
[340]
it's just nice to see
a W for it somewhere.
[343]
Blender, meanwhile,
[344]
doesn't seem to give two hoots
what kind of memory you have
[346]
as long as it's lots and
PugetBench is an interesting case
[350]
because every Adobe app tested here,
[352]
gets a performance uplift,
[354]
but none of them are
as stark as Photoshop.
[356]
Even our low end DDR5 kit is 14% faster
[360]
than our high end DDR4 kit
and that's saying a lot.
[363]
Though, this again makes sense
[365]
because of Photoshop's
memory intensive layer
[367]
and effects operations, where
the GPU isn't really doing
[370]
a lot of the work.
[371]
Finally, we've got SPECworkstation
[372]
where some ridiculously
memory intensive tests
[375]
come into play giving us substantial 20
[377]
and even 30% performance uplifts
in the product development,
[381]
life sciences and energy suites.
[383]
So that stuff like CAD, medical imaging
[385]
and geological surveys
where there's a lot of data
[388]
being thrown around.
[390]
Financial services in
general operations meanwhile,
[392]
not so hot, which we'd expect
considering the lighter nature
[396]
of that work and our previous results.
[398]
Overall then productivity
nets us a rather significant,
[401]
but kind of inconsistent five
to 7% bump in performance,
[406]
DDR5 compared to the fastest DDR4.
[408]
So this is one area where
depending on what you do,
[411]
DDR5 might be worth the investment.
[414]
Although I gotta say, that better be
[415]
some pretty important work you're doing.
[418]
To be clear, none of
what we're saying is that
[421]
DDR5 has no advantages for average gamers.
[424]
It's more power efficient for one.
[426]
And the nature of it's more numerous,
[428]
but smaller burst buffers
means that architectures
[430]
that rely on responsive memory like Ryzen,
[433]
may find more benefit even
at these earlier speeds,
[436]
despite the looser
timings compared to DDR4.
[439]
But since a DDR5 Ryzen
platform doesn't exist yet,
[443]
and by the time it does both
motherboard manufacturers
[446]
and memory manufacturers
might have a better idea
[449]
of how to optimize their implementations,
[451]
I think it's safe to say
that you shouldn't be paying
[453]
a premium for it today.
[455]
Leading us to the big question
we need to answer here,
[457]
why are your scalpers scalping it then?
[459]
Well, I think there's a
big emotional component.
[462]
People like having the
latest and greatest thing.
[465]
And I think that we in
the media need to accept
[468]
at least part of the blame for that.
[470]
But there's another angle to it
[471]
and that's motherboard manufacturers.
[473]
All of the highest end, shiniest,
[475]
most appealing motherboards are DDR5 only.
[479]
And now that there's a shortage of it,
[481]
it's not just as simple for them as,
[483]
plunking down some DDR4 slots
in place of the DDR5 ones
[487]
to improve consumer options.
[489]
Trace layouts and lengths, topology
[491]
and of course the fact that
the power management IC
[494]
is on each DDR5 module rather
than built into the board
[498]
means that there's a non-trivial
amount of work involved
[501]
with retooling a board for DDR4.
[504]
The good news for you today is that,
[506]
most of the highest end
motherboards don't really do
[510]
a lot to justify the extra spend
[513]
at least if you're not
overclocking with liquid nitrogen.
[516]
As we've seen in our previous
motherboard investigation,
[518]
down to earth gaming motherboards like,
[520]
Asus Rg Strix or MSIs Tomahawk series
[523]
are gonna give you
performance and features
[525]
very close to the bleeding edge.
[528]
And if you really must have high-end,
[530]
Gigabyte actually has gone to
the effort of remaking nearly
[534]
their entire lineup in DDR4 flavors,
[537]
leaving only the cream
of the crop out of reach.
[539]
We're gonna have some
links to these down below
[540]
if you're looking to build a 12th gen rig.
[542]
For today, though,
[544]
the solution is to hoist
a great big middle finger
[546]
to scalpers by buying DDR4
while it's still relevant.
[550]
Hopefully by the time you
don't want it anymore,
[552]
the scalpers will have moved
on to greener pastures.
[555]
Maybe some game console
they can buy or something.
[559]
Actually, please don't do that.
[560]
Game console are one of the
only ways people can get a
[563]
reasonably priced gaming rig right now.
[566]
Thanks Brilliant for
sponsoring today's video.
[568]
Brilliant teaches you by
solving puzzles and problems
[570]
on their website and app.
[571]
They've got over 60
interactive courses in math,
[574]
science and computer science,
[575]
and Brilliant is just a great
website to help you achieve
[578]
your goals in STEM.
[579]
Their courses are designed
to puzzle and surprise you.
[582]
And one that probably all of you will like
[584]
is, mathematical fundamentals.
[586]
Is it possible to drag the number of tiles
[588]
so that every row and column adds up
[589]
to the target sum beside it?
[590]
Find out by giving it a try today,
[592]
just go to brilliant.org/linustechtips
to find out.
[595]
The first 200 of you to click on the link,
[597]
will get 20% off today.
[599]
Thanks for watching guys,
[600]
go check out our previous
video on DDR5 for more info
[603]
on the inner workings and
just how it differs from DDR4.
[607]
I promise it is super interesting,
[608]
even if you can't or
shouldn't buy it right now.
Most Recent Videos:
You can go back to the homepage right here: Homepage